Ivan Smirnov

However, now there's this, which seems to have all of the missing paths: (???) ```rust DYLD_FALLBACK_LIBRARY_PATH = " ~/dev/rust/hdf5-rs/target/debug/deps :~/dev/rust/hdf5-rs/target/debug :~/.rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/lib " ```

https://github.com/rust-lang/cargo/pull/6355/commits/b7516d386377dc2de5fd892a5f97bb62cf71888a

> I don't know if this repo is still maintained. Maintained but not actively so. If there's clear issues though, we can investigate and fix.

Thirded.

Agreed re: the doc folder. Re: libqoi and test_gen.rs, I'm unsure about that, if you remove those, you won't be able to simply run `cargo test` because of ``` target/debug/build/libqoi-ec49f992f9130877/out/qoi.c:1:10:...

To be honest, I'm not 100% comfortable with this in general (which also applies to the current version of `.clean_rmeta()`, but at least it only looks in `*/deps` folders...), since...

I think the way the current `clean_rmeta()` works may actually be wrong since it may leave the cache in a half-broken state (with only .rmeta files gone). This PR pretty...

@hannobraun Thanks. Did you actually call `.clean_rmeta()`? (just making sure) Could you please try posting before/after diffs of the `target/` folder? Something like ```sh $ cargo clean $ cargo test...

@hannobraun Thanks, this is helpful. So, I don't see any rmetas like in my case, but rather new rlibs (also, I've tested this on darwin and you're obviously on linux-64)....

To explain the use case, we're using bootstrap-toc in generated/rendered html (a custom Jupyter notebook renderer) which quite often contains over a hundred of headings, thus having a third header...